An Optimization-based Method to Define Dead-band Values of Industrial Alarms

Abstract

Industrial alarms are a way to maintain a proper and secure operation of industrial processes avoiding productive and quality losses, accidents, and possible environmental disasters. However, most alarm systems suffer from the high incidence of chattering alarms, which are alarms that rapidly transit from active and inactive states. To minimize this problem, alarm systems use dead-band techniques to alarm thresholds. However, the alarm dead- band parameters must be well tuned to be effective. In this context, this work proposes a method to improve the quality in specifying alarm dead-bands. The method is based on optimization of performance metrics of alarm systems considering the actual alarm threshold. This method provides effective alarm dead-band values that reduce chattering alarms without being excessively large. To validate our approach, a case study was used to determine the efficiency of the proposed method in comparison with common standards. The data for the case study was generated using Tennessee Eastman Process, which is a well-known simulator in literature.